Observability and the Development Process
summary
Shows how Honeycomb uses observability from day one to choose, build, verify and roll out changes: adds a synthetic field event_time_delta_sec to find client-to-API delays, uses 95th percentile and other quantiles to reveal obvious average-lying patterns, runs parallel simulation with feature flags to see whether new rate-limiter or string-storage logic would have actually limited users, and correlates regressions to build ID to pinpoint which deploy caused latency spikes. Results are concrete and actionable.
AI-written summary · about
chapters (21)
- 0:00Intro
- 4:59Ownership
- 6:18Development Process
- 8:16Building or Fixing
- 8:58Debugging or Reward
- 9:58Example
- 12:57Debug Statements
- 15:28Shared Cache
- 16:03Test Cases
- 17:40Testing in Production
- 19:38Future Flags
- 23:42Watchface
- 25:22Observing Reality
- 28:59Blast Radius
- 29:36Recap
- 30:29Best Practices
- 37:50Migrating from Un unstructured
- 40:54Evolving Tools
- 42:45Visualizing
- 44:36Tracing
- 48:23Conclusion
related talks
31mObservability for Developers: How to Get There
45mObservability is for User Happiness
44mObservability: Superpowers for Developers
27mTradeoffs on the road to Observability
35mObservability: Building Confidence & Answering the What, Where, How
43mObservability for Emerging Infra: What Got You Here Won't Get You There
37mObservability-Driven Development
14mProgressive Delivery with Observability